Impact of Fouling on Flow-Induced Vibration Characteristics in Fluid-Conveying Pipelines

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1

Impact of Fouling on Flow-induced Vibration Characteristics in Fluid-conveying Pipelines

JIANFENG HUANG1,2, Member, IEEE, GUOHUA CHEN1, LEI SHU2, Senior Member, IEEE, YUANFANG CHEN2, Member, IEEE , AND YU ZHANG3 1 Institute of Safety Science and Engineering, South China University of Technology, Guangzhou 510640, China 2 Guangdong Petrochemical Equipment Fault Diagnosis Key Laboratory, Guangdong University of Petrochemical Technology, Maoming 525000, China 3 School of Engineering, University of Lincoln, Lincoln LN6 7TS, U.K Corresponding authors: G. Chen ([email protected]) and L. Shu ([email protected])

 in a non-uniform force field under the influence of various Abstract—This paper addresses monitoring problems excitation forces caused by fluid flow, which easily generates commonly encountered in petrochemical enterprises caused by vibrations [1]. The forming of deposits and fouling are very fouling and clogging in the circulating water heat exchangers by likely when industrial circulation water, the cooling medium of monitoring the heat exchanger’s wall vibration signal for early a heat exchanger, is used. Appropriate vibration is beneficial in failure detection. Due to the difficulties encountered in simulation caused by the large number of tubes inside the heat exchanger, helping to reduce fouling, and serious fouling lowers the heat such methods were discussed by studying in the fluid-conveying exchange efficiency. To maintain its original efficiency, pipeline fouling. ANSYS was used to establish the normal model accelerating the velocity is required, which leads to increased and fouling model of a fluid-conveying pipeline so as to analyze heat exchanger vibration. According to statistics, the changing rule of various parameters that are influenced by approximately 30% of damage to heat exchangers is caused by different inlet velocities. As the inlet velocity and fouling severity bundle vibration[2]. Therefore, the study of vibrations related continuously increased, the wall load and the vibration acceleration increased as well, leading variations in wall vibration to circulation water heat exchanger fouling and clogging is of signals. This paper conduct extensive experiments by using realistic significance. straight pipes to compare the results from simulation and from At present, some researchers have carried out simulation normal fluid-conveying pipelines, under the same working studies on optimizing the heat exchange tube structure, heat conditions. By such comparison, we estimate the accuracy of the exchange performance, and parameters of the heat changer by simulation model. means of ANSYS [3-5]. However, there are few achievements

Index Terms—Fluid-conveying pipelines, fouling impact, on the malfunctioning vibration characteristics of the heat vibration characteristics, vibration signals. exchanger through commercial simulation software, because fluidic and structural two-way coupling is involved, and the heat exchanger is extremely large, leading to difficulties in the I. INTRODUCTION analysis and calculation of a finite element. Hence, this paper HE circulating water heat exchanger is used universally to describes studies conducted on fluid-conveying pipeline Texchange heat, in modern petrochemical enterprises, and fouling and vibration to identify methods that can be used to accounts for 40% of the overall investment in facilities. The diagnose heat exchanger malfunctions caused by vibration maintenance workload accounts for approximately 60–70% of signals. the total maintenance workload. The safe operation of a heat The heat transfer tube is one of the basic components of a exchanger is of great significance to continuous production of circulating water heat exchanger used to transfer and control these enterprises. The fluid movement within the liquid or gas. The number of heat transfer tubes located inside shell-and-tube heat exchanger is extremely complex, and the heat exchanger varies from several to thousands. The includes transverse flow, axial flow, bypass flow, etc. There is a bundles of the shell-and-tube circulating water heat exchanger stagnant zone at both ends of the bundles, where flow velocity are normally straight, and the heat transfer tubes are considered and direction changes are irregular. The heat transfer tubes are a combination of various types of beams. Normally, the pipeline between two baffle plates is a simply supported This work is supported by National Natural Science Foundation of China single-span beam, and the pipeline between the tube plate and (NO.61401107 and NO. 21576102), International and Hong Kong, Macao & baffle plate is similarly a single-span beam with one fixed end Taiwan collaborative innovation platform and major international cooperation and one simply supported end [6]. Therefore, further study of projects of colleges in Guangdong Province (No.2015KGJHZ026), The Natural Science Foundation of Guangdong Province (No.2016A030307029), the Open the vibration characteristics of heat transfer tube fouling of heat Fund of Maoming Study and Development Center of Petrochemical Corrosion exchangers through the vibration characteristic analysis of and Safety Engineering (No.201509A01), the Open Fund of Guangdong fluid-conveying-fouling pipeline is proposed. Provincial Key Laboratory of Petrochemical Equipment Fault Diagnosis (No.GDUPTLAB201605), Guangdong University of Petrochemical The contributions of this paper are listed as follows: Technology through Internal Project 2012RC106. > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 2

1) The bundles of circulating water heat exchanger were displacement of pipe wall points, fixed pivot reacting force, divided into numerous fluid-conveying pipelines. Heat cross-sectional displacement and stress, were analyzed. exchanger pipeline fouling was studied and discussed, resulting ANSYS simulation technology was used by Li [16] to study the in a new diagnostic method. impact of fluid-solid coupling on fluid-filled pipeline vibration 2) A finite element model was put forward to analyze the modality and the impact of variation constraint on fluid-filled fluid-conveying pipeline fouling. By studying the models’ pipeline forces as well as its vibration modality. It was different fouling levels, this paper determined the pattern of discovered that the fluid-solid coupling and increased pipeline vibration caused by pipeline fouling. constraint variation resulted in a rise in pipeline vibration 3) The vibration test experiments were conducted on a frequency and a significant decline in pipeline equivalent stress non-fouling, fluid-conveying pipeline, which proved the and total deflection. Wu [17] and other scholars used a finite correctness of the pipeline finite element model. element method to conduct modal analysis and harmonic The remainder of this paper was organized as follows. The response analysis on high pressure pipe. In the meantime, the related work about pipeline fouling and pipeline vibration was impact on pipeline vibration frequency caused by increased briefly introduced in Section II. Section III described the clamp support and location changes were analyzed. This led to pipeline cell stress analysis and section IV presented the basic the conclusion that the presence of prestress would increase the simulation theories. The pipeline simulation model were built pipeline’s natural vibration frequency. Wang [18] used in section V and section VI. The experimental analysis of fluid-filled pipe for flow field simulation and discovered the normal pipelines was given in Section VII. Finally, Section pressure changing characteristics of the pipe with the periodic VIII presented the overall conclusions. initial velocities. They further combined with the one-way coupling module of an ANSYS workbench, and obtained the II. RELATED WORK fluid-solid coupling vibration modal characteristics of Because fluid-conveying pipeline vibration accelerates fluid-filled pipe under the influence of fluctuating pressure and fatigue damage, shortens the lifespan of, and causes harm to the impact of pipeline support distance on vibration modal. materials (e.g., the connector crack), scholars have carried out Wu [19] presented a simplified model for the numerical extensive research on pipeline vibration. simulation of blockage pipe detection based on the principle of Silva [7-8] presented preliminary research results of auto-oscillation theory. The result shown that the blockage vibrational hammer excitation for easy to use external damping was direct ratio to the flow velocities of pipe and the non-invasive, non-destructive fouling detection in pipelines. location of damping had the cosine relation with damping The proposed method could detect the inner pipe layer parameters so the location and magnitude of blockage could be formation, and thickness estimation of the adsorbed material. fixed if two blockage damping parameters were known. Zhu Then, Silva [9-11] analyzed the vibration signals in presence of [20] put forward the basic continuity equation which could be an inner pipe fouling layer using an accelerometer and a used to calculate the coupled water hammer based on the microphone for detection. The paper outlined the experimental existing theory of water hammer and its coupled theory. Chen setup, achievable sensitivities and limitations of the method. [21] presented an experimental method of non-contact Carellan [12] presented a critical review of the state of the art of vibration measurement to study the parametric resonance of a the approaches based on ultrasonic vibration of the pipe. This pipe conveying fluid. The result showed that at a certain flow paper also analyzed the limitations that were pertinent to the velocity, the phenomena of parametric resonance could occur prevention of fouling in pipes. Ren[13] analyzed the vibration with the proper pulsating amplitude and frequency and the induced from fluid-structure interaction in liquid-filled pipes position of parametric resonance regions were closely related to using the traveling wave method. The vibration characters in flow velocity. Qi [22] built a model of fluid conveying pipe two types of coupled problems such as the junction interaction using finite element pack ANSYS and hydrokinetics pack CFX. in liquid-filled pipes and the pipe vibration stability are The data was exchanged between two packs for fluid –solid analyzed. The thin wall cylindrical tube with both fixed support coupling analysis. The vibration characteristic of the pipe ends was simplified to a beam model by Li [14] and other impacting in a ramp increase velocity was analyzed. The scholars. In addition, the flexural vibration beam theory, vibration respondence of the pipe was calculated in a half sine together with finite element methods, was adopted in producing wave excitation. a natural frequency and vibration type of cylindrical tube, The above research mainly analyzed the vibration around which experiments were designed. By comparing the characteristics of fluid-conveying pipeline without taking the theoretical value with the experimental value, was discovered malfunction caused by pipeline fouling into consideration. The that the fluid causes the decline of the natural frequency of fouling pipelines in exchangers would lead to clogging fault. cylindrical tube. Huang [23] proposed a new fault diagnose method based on Guo [15] established a finite element model of a straight vibration signals and support vector machine. An experimental fluid conveying pipe by ANSYS. The dynamic responses of the model of clogging fault diagnosis in heat exchangers was pipeline with a given input fluid flow velocities pulse, in three studied. The experimental results have shown that the proposed different supported modes, were simulated using the finite method was efficient and had achieved a high accuracy for element method and the method of computational fluid benchmarking vibration signals under both normal and faulty dynamics. The results of pipe vibration, including the conditions. This paper primarily describes the study of the > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 3

influences on vibration caused by fouling in fluid-conveying FFFm3 E pipelines. UUxxx2 1  (3) A CA112 CUE  U D ttttt2  III. CELL STRESS ANALYSIS When the fluctuation velocity surpassed vibration velocity to There are many causes behind vibration in fluid-conveying a large extent, x x could be simplified as UU, and UU pipeline. The three main reasons are as follows [18, 24]: (1) It tt was caused by the poor dynamic balance of the power machines equation (3) could be simplified as: or inappropriate base installation; (2) The vibration was caused U 1 FACUUD（）1 C (4) by pulsating flow; (3) The vibration was caused by the fluid mE11t 2 vortex and air column resonance inside the pipeline. Engineering practices show that, when flexible connection pipes were installed or damping measures were used, the pipeline vibration was primarily caused by pulsating flow [25]. The initiating terminal facilities of power machines, such as a centrifugal pump moved in periodic intermittent motion, led to (A)The analyzed position (B)Stress analysis changes in parameters such as pressure, velocity, density, and flow in the pipeline with variations of time and position. Thus, Figure 2. The cell stress analysis of fouling pipeline pulsating flow was generated. Below we primarily analyze the The inner wall of pipeline would be covered in a thick forces of fluid-conveying pipeline from the perspective of the layer of stemming after fouling. As shown in Figure 2, point A impact of pulsating flow. was randomly picked from the interface of fluid and stemming When unsteady liquid was transported, pulsating flow for stress analysis, and the included angle of F and wall tangent caused pipeline vibration [6]. An assumption was made that the was set as θ. In the meantime, F could be divided into two pipeline vibrated slightly within the elastic range because of forces: F1 in the normal direction, and F2 in the tangential pulsating flow. As shown in Fig.1, point A was randomly direction. Likewise, F1 was decomposed into the axial picked as the junction point of the fluid and the inside wall of a component and the lateral component Fy. The following result normal pipeline to conduct cell stress analysis, and the axial was obtained through the triangle:: FF sin thrust of fluid at point A was set as F . Here, the main lateral 1 (5) force was pressure on the wall generated by pulsating flow, Fm. FF2  cos (6)

FFy  1 cos  F sin cos (7)

At this point, the lateral force was Fm +Fy, which indicated an increase in the fouling pipeline’s lateral, which is one of the reasons it vibrated more violently than the normal pipeline.

(A)The analyzed position (B)Stress analysis IV. SIMULATION THEORY Figure 1. The cell stress analysis of normal pipeline The ANSYS-CFX solution method was used in this paper The pressure Fm was the sum of an inertial force sharing to conduct simulation research on a fluid-conveying pipeline. the same acceleration as the flow and a drag force whose phase To obtain the changing status of vibration acceleration, a was the same as flow velocity. two-way coupling separation solution was adopted. For starters, A structure in accelerated motion was built in the this study used a flow field analysis module to calculate the accelerated flow[26], and its net inertial force at unit length flow field of a fluid-conveying pipeline. Afterward, a structural was: dynamics analysis using a structural analysis module was used 2 UUx to calculate surface pressure results and surface velocity results. FA31 CA 2 (1) ttt The structural distortion was transferred to CFX to conduct The drag force at structural unit length was: flow field analysis. Such an approach maximized the use of 1 xx structural mechanics and fluid mechanics calculations. FEECU U D1 (2) 2 tt According to the given order, the solid control equation and In this equation, ρ is set as the fluid density, A is the fluid control equation was solved. In order to obtain more 2 structural cross section (m ), C1 is the added mass coefficient, accurate simulation results, the results of solid and fluid control U is the mean velocity of transient flowing fluid cross section were transferred and exchanged through the contact surface.

(m/s), CE is the drag force coefficient, and D1 as the width of the In terms of fluid-solid coupling, seeking solutions to a structure approaching the flow section. In addition, the structural dynamics equation, Navier-Stokes Equations, and a direction of drag force was consistent with the direction of the fluid continuity equation at the same time is required [27]: x The structural dynamics equation was: relative velocity of both fluid and structure U  . t [M] + [C] +[K] = F (8) Therefore, the resultant force was: In this equation, [M] represented the overall structural mass matrix; [K] represented the overall structural stiffness > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 4 matrix; [C] represented the overall structural damping matrix; parameters were as shown in Table I. In order to be consistent [F] represented the structural load matrix, and {u} represented with the following experiments, the flow velocities at the the structural displacement vector. entrance point of the fluid-conveying pipelines were set The fluid momentum equation was: respectively as 0.33, 0.52, 0.74, and 1.01 m/s. Moreover, the + ν(ν ∙∇) =− （∇ + divτ） (9) mesh used in this model was divided into two parts: the ρ structural mesh and the fluid domain mesh. The former was The fluid continuous equation was: divided into 5,795 units and 35,158 nodes, while the latter was ρ +∇∙(ρν) =0 (10) 43,946 units and 38,250 nodes. Where ∇ was Hamiltonian, t was time ， p was fluid TABLE I BASIC PARAMETERS OF PIPELINE MODEL pressure, was fluid density, was fluid velocity, τ was Item Parameter Item Parameter shearing force. This paper did not take the energy equation into length 2000mm yield strength σ=0.25GPa consideration, because energy transfer and conversion was not elasticity involved. modulus E =20GPa Poisson's ratio υ=0.3 The turbulence model was used to conduct analysis on the environment 3 3 ℃ simulation of flow field in the fluid-conveying pipeline. The density 7.85×10 kg/m temperature 25 fluid density standard k- model was adopted in the CFX [28], which was （water） ρ=1.0×103kg/m3 pipe diameter φ20×2mm composed by the turbulent kinetic energy equation k and the diffusion equation . 2) Boundary condition setting The turbulent kinetic energy equation k was: First, the boundary condition of the structural body was set. ( ) The inner wall surface was set as the interface of fluid-solid = + + − (11) coupling. In addition, the wall surface at both pipeline ends was The diffusion equation was: set as the constraint condition of securing support, the gravity ( ) 2 ( ) field was set as 9.8 m/s with the direction of the −Y axis, and = + + − (12) the inner wall surface was considered to be the loading surface Where = ′′ ，= of fluid pressure load. Second, the boundary condition of the fluid domain was Turbulent viscosity expressed as function of k and : set. Transient analysis was adopted, because the bidirectional = (13) process was transient transfer, and the duration and step length represented the pressure-generating item caused by analyzed should be the same as those in structural analysis. velocity gradient: Dynamic mesh was chosen for the fluid domain, the heat

transfer was shut down, and the k-epsilon turbulence model = + (14) was used. The inlet flow velocities were set as fixed values. The Where =0.09， =1.44， =1.92, = pressure at exit points was set at 0 Pa, and the boundary of wall 1.0， = 1.3 surface was considered to be the fluid-solid coupling surface. The boundary condition of no-slip from integral to wall The total time length of solution was set as 1 s, the step length surface was = 0 and = 0. was set at 0.01 s, and appropriate condition of convergence was Starting from the turbulence model, the following portion set in the CFX solver. of the study involved creating a normal fluid-conveying B. Pipeline Fluid Domain pipeline model without fouling by means of ANSYS to study The flow velocity nephogram of fluid domain was extracted the vibrations of pipeline under the influences of different flow to observe the flow velocity distribution and flow direction in velocities. Afterward, a fluid-conveying pipeline models were the fluid-conveying pipeline. Figure 3, shows that in the built with different fouling levels that were used to conduct pipeline model with different input flow velocities, the high research on the fouling pipeline vibrations with the premise of flow velocity intensively distributed at the center of the pipeline same flow velocity. when the fluid was running through. On the contrary, the flow velocity near the wall surface was low, and the flow velocities V. FINITE ELEMENT ANALYSIS ON NORMAL at the entry points were lower than those inside the pipeline. In FLUID-CONVEYING PIPELINE addition, the direction of fluid was consistent, moving in A. Model and Boundary Conditions toward the outlet in a rectilinear way. However, due to the 1) Basic model parameters uneven distribution of flow velocities inside the pipeline, there This paper established a finite element model of were velocity differences, which caused changes in pipeline fluid-conveying pipeline and a finite model of fluid inside the load to a certain extent, which in turn generated pipeline pipeline. Galvanized steel pipeline was chosen, and specific vibration. > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 5

Figure 3. Flow velocity analysis of a pipeline relatively large at the wall of the inlet, while the inlet flow C. Pipeline Wall Load velocity was small. The differences in flow velocity at the inlet As shown by the extracted pipeline wall load in Figure 4, the had a significant impact on the pipeline load, which increased load showed a gradient decrement trend from inlet to outlet. with the growth of inlet flow velocity. According to the above The closer it was to the inlet, the larger the pipeline wall load analysis on the flow velocities inside the pipeline, it is apparent became, while the closer it was to the outlet, the smaller the that the flow velocity differences in axial direction at the inlet load became. With the continuous growth of inlet flow velocity, are relatively large. Thus, it is concluded that flow velocity the wall load successively increased with the same load difference increase is one of the reasons for the larger load at direction, radially covering the surface of pipeline wall. the pipeline inlet. Combined with Figure 3, it was discovered that the load was

Figure 4. Wall load analysis of a pipeline

Figure 5. The displacement analysis of a pipeline Figure 6. The acceleration analysis of a pipeline D. Wall Displacement and Acceleration the five flow velocities in the computer model, the pipeline The nephograms of pipeline wall displacement and deformation was basically consistent. The relatively larger acceleration were extracted as shown in Figures 5 and 6. With deformation appeared at the center of the pipeline, while the > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 6 relatively small deformation displacement symmetrically displacement values influenced by various flow velocities were appeared at both ends of the pipeline, which was relevant to the as shown in table II. It was discovered that the average constraints of pipeline model. Because the acceleration and displacement values at these three observation points were displacement were quadratic integrals, the nephogram of wall different with different flow velocities. Flow velocity values displacement and acceleration nephogram showed a consistent are higher at the center and smaller at both ends. Moreover, changing trend. displacement values increased consistently with the growth of Three positions, A (near inlet), B (near the center) and C flow velocity. Therefore, flow velocity is one of the main (near outlet), were selected as the observation points on the influences on vibration in fluid-conveying pipeline. same axis of the fluid-conveying pipeline. The average displacement values influenced by various flow velocities were TABLE II THE AVERAGE DISPLACEMENTS OF EACH OBSERVATION POINT as shown in table II. It was discovered that the average Inlet flow Position A Position B Position C displacement values at these three observation points were velocity different with different flow velocities. Flow velocity values -6 -6 -6 are higher at the center and smaller at both ends. Moreover, 0.23m/s -0.6×10 m -24.8×10 m 0.5×10 m displacement values increased consistently with the growth of 0.33m/s -0.3×10-6m 7.0×10-6m 1.2×10-6m flow velocity. Therefore, flow velocity is one of the main -6 -6 -6 influences on vibration in fluid-conveying pipeline. 0.52m/s 0.3×10 m 70.9×10 m 2.5×10 m Three positions, A (near inlet), B (near the center) and C 0.74m/s 0.9×10-6m 146.0×10-6m 4.1×10-6m (near outlet), were selected as the observation points on the -6 -6 -6 same axis of the fluid-conveying pipeline. The average 1.01m/s 1.8×10 m 238×10 m 6.0×10 m pipeline changed with the stemming shape, and high flow VI. FINITE ELEMENT ANALYSIS ON FOULING velocities appeared in the positions where relatively intensive FLUID-CONVEYING PIPELINE fouling exist. The distribution of the flow velocity of fluid A. Model and Boundary Conditions domain inside the fouling pipeline was as follows: The positions near the pipeline wall distributed low flow velocity, while the center showed high flow velocity; at the locations where the stemming was heavily distributed, the direction of fluid changed suddenly, and the flow velocity increased drastically. C. Pipeline Wall Load Analysis As shown in Figure 9, the distribution rules of wall surface load in the four conditions were consistent, which was a gradient decrement trend from inlet to outlet. The closer it was to the inlet, the larger the wall load. The nearer it was to the outlet, the smaller the wall load. For the normal fluid-conveying pipeline, the wall load was evenly distributed Figure 7. The side elevation of the fouling pipeline model with a grading gradient decrement trend. The pipeline in The basic parameters and boundary conditions of fouling fouling condition showed a linear gradient decrement. Due to fluid-conveying pipeline models were consistent with the the roughness of the wall surface caused by stemming, the above-mentioned fluid-conveying pipeline model under normal pressure direction changed with different surface conditions. conditions. There were four levels of fouling: 0%, 20%, 40%, Therefore, the wall load direction of the pipeline with and 60% fouling. The stemming material was the same as that stemming was scattered. Judging from the wall load, its of the fluid-conveying pipeline. The percentages refer to the maximum value constantly increased from 1.7 × 106 to 1.7 × proportion of stemming inside the pipeline, and accounts for 107 Pa, which indicated that the wall load increased with the the whole volume of an empty pipeline. For the purposes of severity of fouling. As shown in Figure 10, there was a clear emphasizing the impact of fouling on vibration, the fouling difference between the fluid domain and the wall loads in levels were set at increments of 20%. As shown in the fouling normal pipeline and fouling pipeline. Hence, the fouling of the model (Figure 7), the shapes of inner wall stemming were fluid-conveying pipeline was the main factor influencing the formed by rotating the random curve at 360°, and the flow pipeline vibration. With the same inlet flow velocity as the velocity at the inlet was uniformly divided into two conditions: premise, the more severe the pipeline fouling, the more violent 0.4 m/s and 1.0 m/s. the pipeline vibration. B. Pipeline Fluid Domain D. Wall Displacement and Acceleration The flow velocities of fluid domain inside the fouling By comparing the wall displacement shown in Figure 11 pipeline are shown in Figure 8. As opposed to the nephogram with the flow velocity of fluid domain shown in Figure 8, it is with evenly distributed flow velocities of fluid domain without clear that the flow velocity increased where stemming was fouling, the flow velocities of fluid domain inside the fouling > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 7 heavily distributed, and the direction change became more obvious with increased wall displacement.

Figure 8. Flow velocity analysis of the fouling pipeline

Figure 9. Wall load analysis of the fouling pipeline

Figure 10. The contrast of the fluid domain and wall loads

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 8

Based on the average displacement value acquired in the was discovered that the distribution rule of acceleration was steady condition of wall surface as shown in table III, it was similar to displacement. In regions with heavily distributed discovered that the displacements were all different under the stemming, the acceleration increased, and the flow direction different flow velocity circumstances and different degrees of changed more rapidly. As a consequence, the force on the fouling. With the same inlet flow velocity, the pipeline pipeline wall increased as well, ultimately resulting in larger displacement increased with the severity of fouling. While the wall displacement and acceleration. displacement of pipeline with higher flow, velocity was higher The acceleration in four fouling conditions and two different in the same fouling condition, but otherwise gradually reduced. inlet flow velocities is summarized in Figure 13. In cases where Therefore, with the premise of same inlet flow velocity, the 0% or 20% fouling exists, the variation of acceleration was displacement of pipeline in a steady condition was considered insignificant regardless of whether the inlet flow velocity was to be one of the parameters used to monitor and diagnose 1.0 or 0.4 m/s. With 40% or 60% fouling, the acceleration pipeline malfunction. However, in practical situations, an changed substantially regardless of whether flow velocity was acceleration sensor was normally used to test the vibration of 1.0 or 0.4 m/s. The above results showed that only when the pipeline. It is of more practical significance to discuss the fouling was severe enough would it cause significant pipeline acceleration nephogram. vibration. In the cases of 0%, 20%, and 40% fouling, the wall TABLE III DISPLACEMENTS CONTRAST OF DN20 PIPELINE (UNIT:m) acceleration increased with the growth of flow velocity. flow no 20% 40% 60% However, with 60% fouling, the acceleration with a flow velocities fouling fouling fouling fouling velocity of 0.4 m/s was similar to that of 1.0 m/s. Therefore, the 0.4m/s 0.0066 0.0075 0.029 0.047 inlet flow velocity and fouling severity were the main 1.0m/s 0.0078 0.0087 0.038 0.051 influencing factors in fluid-conveying pipeline vibration. The wall acceleration nephogram is shown in Figure 12. It

Figure 11. The displacement analysis of the fouling pipeline Figure 12. The acceleration analysis of the fouling pipeline

Figure 13. The acceleration contrast on the pipeline in different flow velocities the ANSYS and CFX to establish the finite element model of VII. EXPERIMENTS ON NORMAL PIPELINE fluid-conveying pipeline, this paper describes how experimental analysis was carried out on the experiment model A. Experiments under a single working condition. Using a fluid-conveying In order to verify the feasibility and the correctness of using pipeline without fouling as a research subject, the experimental > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 9 facilities were as shown in Figure 14 and Figure 15. The The experiment primarily aimed to verify the consistency of facilities consisted of simple devices used to test the vibrations the simulation and experimental model data of fluid-conveying of pipeline with two pipe diameters under different flow pipeline. Hence, the acceleration data of vibration in the central velocity conditions. It included a wireless vibration signal test portion of the pipeline was used for analysis—especially the system, an experiment pool, pipeline located outside the pool, a acceleration value under different flow velocities at this stop valve at the front end of the pipeline, a pipeline pressure position. Multiple high precision wireless sensors were test instrument, flexible pipeline used to connect the pipeline installed at the center and at both ends of the pipeline through and pool, a booster pump the at the pipeline inlet, and flow two fixed supports to collect vibration signals at the same time. meters at both ends of the pipeline. The basic parameters of the Tap water was used for low flow to avoid the influences caused experimental pipeline were as shown in Table IV. This paper by pump vibration. Under circumstance where a demand for conducted experiments respectively on the pipeline, using two high flow velocity existed and the tap water alone failed to different pipe diameters to simulate the vibrations caused by provide sufficient water, the pump was initiated and used to fluid with different inlet flow velocities. The pipeline inlet was supply water. The flow velocity was set at 0.23, 0.33, 0.52, 0.74, connected to the pool and tap water pipeline. In addition, the and 1.01 m/s to respectively measure the acceleration of pool was used to provide a water supply to the experimental pipeline vibration under the influence of different flow pipeline through a water pump. The tap water increased the velocities. flow of pipeline to further expand the measuring range of the B. Data Comparison experiment. Flexible pipeline was used to connect the pump TABLE V THE MAXIMUM ACCELERATIONS CONTRAST OF PIPELINES and pipeline to reduce the impact of the pump vibration on the DN20 DN20 DN25 DN25 whole experiment, since pump vibrations directly influence Flow velocities simulation experiment simulation experiment signal collection. (m/s) （g）（g）（g）（g） TABLE 4 THE BASIC PARAMETERS OF PIPELINES 0.23 0.0588 0.0736 0.0160 0.0546 Type Material Length Pipeline diameter 0.33 0.0701 0.0848 0.0340 0.0615 DN20 galvanized pipe 2000mm φ20×2mm DN25 galvanized pipe 2000mm φ25×2mm 0.52 0.1071 0.0864 0.0509 0.0610 0.74 0.2300 0.0728 0.0579 0.0933 1.01 0.3815 0.0905 0.0772 0.0601

Using the vibration data at the center of the DN20 pipeline as an example, Figure 16 shows that the data obtained by finite element analysis and experiment were generally similar with a velocity of 0.52 or 0.74 m/s. The acceleration data at a velocity of 0.33 m/s was within the same order of magnitude, while there was an order of magnitude difference at the velocity of 1.01 m/s. Meanwhile, the maximum vibration acceleration with different flow velocities was extracted. As shown in Table V, it can be seen that the maximum acceleration simulated by computer increased with the growth of inlet flow velocity with Figure 14. Schematic diagram of straight pipeline experiments:1-wireless an extraordinarily clear linear growth trend. In addition, the vibration sensors, 2-flowmeters, 3-pumb, 4- water valve, 5- flexible pipeline, 6-pool, 7-press instrument, 8- tap water inlet, 9-pipelines. experiment data also showed such a trend with a lower linear growth. When the inlet flow velocity was above 0.5 m/s, the maximum acceleration of the two models gradually deviated from each other. The data of the central portion of the DN25 pipeline showed basically the same trend. As shown in Figure 17, with inlet velocities of 0.52, 0.74, and 1.01 m/s, the main acceleration data were mostly the same. At a velocity of 0.33 m/s, the accelerations were within the same order of magnitude. Moreover, the maximum acceleration increased continuously with the growth of flow velocity. By comparing several vibration acceleration experiments with the above five types flow velocities, it was discovered that the data simulated by computer and the experimental data were basically within the same order of magnitude range. Considering the influence of the water pump’s vibration, instrumental error, boundary conditions of finite element Figure 15. Vibration test experiment of pipelines analysis, and the differences in mesh quality, the results are > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 10 believed to be reasonable. In other words, it is acceptable to use fluid-conveying pipeline vibration to acquire the acceleration ANSYS and CFX two-way coupling methods to simulate the of the pipeline surface.

Figure 16. The acceleration contrast of DN20 pipeline

Figure 17. The acceleration contrast of DN25 pipeline inlet flow velocity. Therefore, the inlet flow velocity of VIII. CONCLUSION fluid-conveying pipeline was one of influences over vibration. The pulsating flow triggered the pipeline vibration. After 2) By building the finite element model of fluid-conveying analyzing the forces of fluid-conveying pipeline both before pipeline with different fouling levels, this paper conducted and after fouling, it is believed that one of the causes of analysis on the flow velocity, direction, pipeline wall load, wall aggravated pipeline vibration is the increase of lateral force. displacement, and acceleration of fluid domain in a ANSYS finite element analysis was used to establish the finite fluid-conveying pipeline under the influence of four different element model of fluid-conveying pipeline under the influence levels of fouling (0%, 20%, 40%, and 60% fouling). It was of fluid-solid coupling. Afterward, this paper conducted finite discovered that the flow velocity and direction of fluid domain element analysis on the pipeline with different flow velocities changed according to the variation of fouling positions. In and fouling levels. Furthermore, a group of pipeline vibration addition, the wall load, wall displacement, and acceleration experiments were studied under the same normal condition. It increased with increased degrees of fouling. Hence, the found that the simulation data and the experimental data were fluid-conveying pipeline fouling also had an impact on the basically in accordance. pipeline vibration. 3) Two different experimental research studies were carried The main conclusions were listed as follows: out on fluid-conveying pipeline, and with two different pipe 1) By establishing the finite element model of fluid-conveying diameters. This paper proved the correctness of the simulation pipeline under normal conditions, this paper analyzed the flow model by comparing the simulation data with experiment data velocity, direction, pipeline wall load, wall displacement and of non-fouling, fluid-conveying pipeline under the same the acceleration of fluid domain in fluid-conveying pipeline of working conditions. four cases. It was discovered that the wall load, wall displacement, and acceleration increased with the increase of > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 11

REFERENCES [24] W. Song, T. Xiao, J. Li. “On the present status and prospect of pipeline vibration control technology,” (in Chinese), Journal of Safety and [1] A. Li, G. Xin, L. Zhou. “Research state on fluid-induced vibration of heat Environment. 2012(03): 184-188. exchanger,” (in Chinese), Liaoning Chemical Industry. 2007(12): [25] J. Qian. “Study on noise and vibration reduction of providing water 849-852. system to high building,” (in Chinese), M.S. thesis, Tianjin University Of [2] Y. Lai, M. Liu, Q. Dong. “Dynamic characteristic analysis of heat Science & technology, Tianjin, China, 2002. exchanger tube bundles with external fluid,” (in Chinese), Pressure [26] C. Yang. “Vibration characteristics of unsteady-fluid-filled pipe system Vessel Technology. 2004(12): 22-25. and vibration restrain,” (in Chinese), PHD thesis, Huazhong University of [3] A.K. Tiwari, P. Ghosh, J. Sarkar, H. Dahiya, and J. Parekh. “Numerical Science and Technology, Hubei, China, 2007. investigation of heat transfer and fluid flow in plate heat exchanger using [27] Y. Lai. “Analysis on dynamic characteristic and flow induced vibration nanofluids,” International Journal of Thermal Sciences. 2014, 85(0): of heat exchanger tube bundles,” (in Chinese), PHD thesis, Nanjing Tech 93-103. University, Jiangsu, China, 2006. [4] W. Du, H. Wang, and L. Cheng. “Effects of shape and quantity of helical [28] V. Yakhot, SA. Orszag, S. Thangam, TB. Gatski, and CG. Speziale. baffle on the shell-side heat transfer and flow performance of heat “Development of turbulence models for shear flows by a double exchangers,” Chinese Journal of Chemical Engineering. 2014, 22(3): expansion technique,” Physics of Fluids A (Fluid Dynamics). 1992, 4(7): 243-251. 1510-1520. [5] H. Bilirgen, S. Dunbar, EK. Levy. “Numerical modeling of finned heat exchangers,” Applied Thermal Engineering. 2013, 61(2): 278-288. [6] L. Cheng. “The flow induced vibration in heat exchanger,” (in Chinese), China, Beijing: Science Press, 1995. [7] JJD. Silva, AMN. Lima, FH. Neff, and JSD. Neto. “Fouling detection JIANFENG HUANG(M'15) received the based on vibration analysis with the hammer impact test,” Warsaw, master's degree from the Guangdong Poland: Institute of Electrical and Electronics Engineers Inc., 2007. University of Technology, Guangzhou, [8] JJD. Silva, AMN. Lima, FH. Neff, and JSD. Neto. “Hammer impact test applied for fouling detection in pipelines,” Sba Controle & Automao China, in 2006. He is currently pursuing the Sociedade Brasileira De Automatica. 2011, 22(6): 620-630. Ph.D. degree with the Institute of Safety [9] JJD. Silva, IB. Queiroz, AMN. Lima, and JSD. Neto. “Vibration analysis Science Engineering, South China for fouling detection using hammer impact test and finite element University of Technology, Guangzhou. simulation,” Victoria, BC, Canada: Institute of Electrical and Electronics Engineers Inc., 2008. Since 2006, he has been with the [10] JJD. Silva, AMN. Lima, FH. Neff, and JSD. Neto. “Non-invasive fast Guangdong University of Petrochemical Technology. He is detection of internal fouling layers in tubes and ducts by acoustic currently an Associate Professor and the Department Head of vibration analysis,” Institute of Electrical and Electronics Engineers Inc., Safety Engineering. Since 2013, he has been with the 2009. [11] JJD. Silva, AMN. Lima, FH. Neff, and JSD. Neto. “Vibration analysis Guangdong Petrochemical Equipment Fault Diagnosis Key based on hammer impact test for multilayer fouling detection,” Lisbon, Laboratory. His research interests include industrial wireless Portugal: IMEKO-International Measurement Federation Secretariat, sensor networks, petrochemical equipment fault diagnosis, and 2009. safety assessment. [12] IGD. Carellan, P. Catton, C. Selcuk, and TH. Gan. “Methods for detection and cleaning of fouling in pipelines,” Ioannina, Greece: Taylor and Francis Inc., 2012. [13] J. Ren, L. Lin, S. Jiang. “Traveling wave method for pipe fluid-structure GUOHUA CHEN received the Ph.D. interaction,” (in Chinese), Chinese Journal of Applied Mechanics. degree in process safety from Nanjing 2005(04): 530-535. [14] B. Li, L. Xie, X. Guo, and Y. Wei. “Effect of flowing fluid on vibration Tech University, China. He is the frequencies of a thin-walled cylindrical tube,” (in Chinese), Journal of Councilor of the China Occupational Vibration and Shock. 2010(07): 193-195. Safety and Health Association. He is [15] Q. Guo, Q. Fan. “Simulation of Fluid-structure Interaction in Fluid currently a Professor and the Department Conveying Pipes by ANSYS,” (in Chinese), Microcomputer Applications. 2010(04): 9-11. Head of the Institute of Safety Science [16] S. Li, B. Lei, Y. Li. “Modal analysis of pipe vibration under liquid-solid Engineering with the South China coupling condition,” (in Chinese), China Metalforming Equipment & University of Technology, Guangzhou, China. His research Manufacturing Technology. 2012(04): 76-78. interests include safety, reliability, and risk assessment [17] Y. Wu, Y. Li, and Z. Liu. “the vibration analysis in full liquid pipe,”(in Chinese), Journal of Vibration, Measurement & Diagnosis. 2013(S2): technology about process equipment. 100-104. [18] F. Wang, Z. Xu, W. Wu, and S. Liu. “Dynamic characteristic analysis of LEI SHU (SM'16) received the Ph.D. fluid-filled pipe under the action of fluctuating pressure,” (in Chinese), degree from the National University of Noise and Vibration Control. 2013(02): 11-14. [19] Y. Wu, Y. Xu, F. “Wang. Blockage detection in water-supply pipelines Ireland, Galway, Ireland, in 2010. Until based on auto-oscillation theory,” (in Chinese), Journal of Harbin 2012, he was a Specially Assigned Institute of Technology. 2014(08): 45-50. Researcher with the Department of [20] J. Zhu. “Improvement of axial-coupled water hammer vibration Multimedia Engineering, Graduate School equation,” M.S. thesis, Kunming University of Science and Technology, Yunnan, China, 2014. of Information Science and Technology, [21] B. Chen, M. Deng, J. Zhang, and Z. Yi. “Experimental study on Osaka University, Japan. He is a member of parametric resonance of pipe conveying fluid based on laser vibration the IEEE ComSoc, EAI, and ACM. Since 2012, he has been measurement technique,” (in Chinese), Journal of Vibration, with the Guangdong University of Petrochemical Technology, Measurement & Diagnosis. 2014(03): 560-564. [22] H. Ji, N. Jiang, X. Huang. “Fluid-solid coupling analysis of fluid China, as a Full Professor. He is the Founder of the Industrial conveying pipes with finite element method,” (in Chinese), Mechanical Security and WSNs Laboratory. He is serving as the Engineer. 2016(01): 205-207. Editor-in-Chief of EAI Endorsed Transactions on Industrial [23] J. Huang, G. Chen, S. Lei, S. Wang and Z. Yu. “An experimental study of Networks and Intelligent Systems, and an Associate Editor for clogging fault diagnosis in heat exchangers based on vibration signals,” IEEE Access, 2016( 4): 1800-1809. a number of famous international journals. He served as more > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 12 than 50 various co-chairs for international conferences and YU ZHANG received the Ph.D. degree workshops. His research interests include WSNs, multimedia from the Department of Civil communication, middleware, security, and fault diagnosis. Engineering, University of Nottingham, Nottingham, U.K., in 2011. She is Yuanfang Chen (M'09) received her currently a Senior Lecturer with the Ph.D. degree and M.S. degree from Dalian School of Engineering, University of University of Technology, China. She Lincoln, Lincoln, U.K. Her research currently works in Guangdong University interests include intelligent systems and of Petrochemical Technology as a machine fault detection and diagnosis. Specially Assigned Researcher. She was an assistant researcher of Illinois Institute of Technology, U.S.A. with Prof. Xiang-Yang Li, from 2009 to 2010. She had been awarded the MASS 2009, 2013 IEEE Travel Grant, SIGCOMM 2013 Travel Grant, IWCMC 2009 and MSN 2010 Invited Paper. She has served as volunteer of MobiCom & MobiHoc 2010 and IEA-AIE 2012. She has been invited as the Session Chair of some good conferences, e.g., MobiQuitous 2013, ICA3PP 2015, and the TPC member, e.g., Globecom 2014, MobiApps 2014. She has served as reviewer of some top journals and conferences, e.g., IEEE TII, IEEE TPDS and Ad Hoc & Sensor Wireless Networks. She is an associate editor of EAI Endrosed Transactions on Industrial Networks and Intelligent Systems.